Allow nested seed packets #2

jkomoros · 2023-06-17T21:57:08Z

A SeedPacket where the seeds have instead of SeedRef, a SeedData, which is expanded to be a topline seed.

Add a id field to seeds (which is validated to verify it matches the local ID) since some seeds will otherwise have implied IDs. If it's not filled in then it will be filled in with a generic and deterministic ID so you don't HAVE to create one
expandSeedData should actually unroll seeds
Add tests for nested data (and all ID auto-generating behavior)
Clean up naming (e.g. Nested for nested types, versus the current of empty for input and Expanded (confusingly) for ones that aren't nested)
Document

The text was updated successfully, but these errors were encountered:

…tch the seed it's planted with. Part of #2.

Currently they're the same exact types, just aliases. The idea is that the non-expanded are the raw packet, and might include nested seeds. Part of #2.

…ough, but will soon allow nested seeds. Part of #2.

jkomoros · 2023-06-24T13:22:47Z

This is being developed on branch nested-seeds. It's really finicky to do automatic type squanching to not repeat yourself with the seed data types (and also they need to be self-referencing!)

jkomoros · 2023-06-24T13:22:51Z

First look at how zod.required works.

Replicate that with a thing that puts a distinctive sentinel in place every time it finds a seedReference.

Then process the union of seed data and make the nested one too automatically

Then do the lazy/recursive thing. Make expandSeedData throw if it finds one that doesn’t work.

Then make expandSeedData work with nested data (if id is set, use that. If not, use the id from parent plus a dash and the parent peppery name

jkomoros · 2023-06-24T13:33:05Z

https://stackoverflow.com/questions/76008929/is-there-a-way-in-zod-to-create-a-new-schema-from-an-object-schema-by-type

https://stackoverflow.com/questions/76293159/how-to-iterate-over-properties-and-get-the-types-in-zod

It's a no op in terms of typing. Having a helper like this makes it a step more likely to automatically factor out making nested versions. Part of #2.

…shape. No semantic difference yet, but getting more of the type construction automated. Part of #2.

…alue automatically. No semantic difference, just the type construction for seed data is automated. There's a cast in there (as Shape) that doesn't feel right, but it still works so 🤷 Part of #2.

Part of #2.

Everything builds and runs, but if you try to grow a sub-seed that was nested, it will fail with an invalid id warning because expandSeedData doesn't yet know how to properly unroll nested seeds. Part of #2.

Part of #2.

…ypes. makeNestedSeedData is harder to do because of the recursive typing. Part of #2.

Just do the same typescript rewriting as done for makeSeedData. Left a comment in a few of the places where the typescript type is wrong to remind myself. Part of #2.

This will help enumerate through keys in expandSeedData. Part of #2.

…ted seed. For now it just barfs with an error. Note this means that tests currently pass, but if you run `garden` it will bail because `default.json` has a nested seed, so the seed packet will fail to parse. Part of #2.

…if set. Part of #2.

Actually, I don't know why this works, but it does seem to work in the `garden` command at least for `nested-example`. Part of #2.

Part of #2.

It requires an ugly type cast. 🤷 Part of #2.

Part of #2.

One doesn't yet pass so commented out. Part of #2.

Before, the seedRference would notice things with an 'id' and match first. But we want to check for the more restrictive seedData first, because if seedReference is matched first we won't find named sub-seeds, they'll be removed by the seedPacket.parse. Part of #2.

Part of #2.

Part of #2. Merge branch 'nested-seeds' * nested-seeds: (25 commits) Document nested seeds in README. Add another matching test for behavior from two commits ago. Add a test that now behaves correctly from last commit. Change order of union of types in makeNestedSeedReferenceProperty. Add a bunch of expandSeedPacket tests. Add a basic test for nested examples. Document the makeNestedSeedData problem, pointing at issue #16. In expandSeedData, leave behind nested seeds with proper SeedReference. Make expandSeedID return the ID it actually chose. Made expandSeedData correctly recurse into nested seeds and unroll them. Made it so expandSeedData will use the override id set within a seed if set. expandSeedData recurses through items and detects when it finds a nested seed. Create SeedDataReservedKeys. Do a hack of a fix in makeNestedSeedData. Fix typescript typing of makeSeedData to have the proper typescript types. Add a TODO based on a bug I just realized in typing. Clean up proper types of SeedData and expandSeedData. Redefine SeedPacket to actually allow nesting of any seed reference. Unexport the seedData* Zod types, since we'll have a larger number now. Make it so makeSeedData runs the makeSeedReferenceProperty on every value automatically. ...

Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Make it so seedData can provide an optional "id", that if set must ma…

9c0f8a0

…tch the seed it's planted with. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Introduce the notion of ExpandedSeed{Packet,Data} and non-expanded.

3b0e423

Currently they're the same exact types, just aliases. The idea is that the non-expanded are the raw packet, and might include nested seeds. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Introduce expandSeedPacket(). It's at this point basically a pass-thr…

ee17283

…ough, but will soon allow nested seeds. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Factor out makeSeedReferenceProperty as a helper for making seedData.

7002c7c

It's a no op in terms of typing. Having a helper like this makes it a step more likely to automatically factor out making nested versions. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Add makeSeedData, which takes a configuration object and returns the …

ac6b26e

…shape. No semantic difference yet, but getting more of the type construction automated. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Unexport the seedData* Zod types, since we'll have a larger number now.

56648a5

Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Clean up proper types of SeedData and expandSeedData.

0659247

Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Add a TODO based on a bug I just realized in typing.

05e73d1

Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Fix typescript typing of makeSeedData to have the proper typescript t…

2b9f2a4

…ypes. makeNestedSeedData is harder to do because of the recursive typing. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Do a hack of a fix in makeNestedSeedData.

9478250

Just do the same typescript rewriting as done for makeSeedData. Left a comment in a few of the places where the typescript type is wrong to remind myself. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Create SeedDataReservedKeys.

64b3aa7

This will help enumerate through keys in expandSeedData. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Made it so expandSeedData will use the override id set within a seed …

abb20ff

…if set. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Made expandSeedData correctly recurse into nested seeds and unroll them.

d367432

Actually, I don't know why this works, but it does seem to work in the `garden` command at least for `nested-example`. Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

Make expandSeedID return the ID it actually chose.

b5420d0

Part of #2.

jkomoros added a commit that referenced this issue Jun 24, 2023

In expandSeedData, leave behind nested seeds with proper SeedReference.

017386d

It requires an ugly type cast. 🤷 Part of #2.

jkomoros mentioned this issue Jun 25, 2023

Fix the nested SeedData type definition problem #16

Open

jkomoros added a commit that referenced this issue Jun 25, 2023

Add a basic test for nested examples.

df9725b

Part of #2.

jkomoros added a commit that referenced this issue Jun 25, 2023

Add a bunch of expandSeedPacket tests.

b21a9d4

One doesn't yet pass so commented out. Part of #2.

jkomoros added a commit that referenced this issue Jun 25, 2023

Add a test that now behaves correctly from last commit.

84d154e

Part of #2.

jkomoros added a commit that referenced this issue Jun 25, 2023

Add another matching test for behavior from two commits ago.

3969030

Part of #2.

jkomoros added a commit that referenced this issue Jun 25, 2023

Document nested seeds in README.

ec5951e

Part of #2.

jkomoros added a commit that referenced this issue Jul 1, 2023

Throw if a seed with a duplicate ID is defined in a packet.

9cae9d8

Part of #2.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Allow nested seed packets #2

Allow nested seed packets #2

jkomoros commented Jun 17, 2023 •

edited

Loading

jkomoros commented Jun 24, 2023

jkomoros commented Jun 24, 2023

jkomoros commented Jun 24, 2023 •

edited

Loading

Allow nested seed packets #2

Allow nested seed packets #2

Comments

jkomoros commented Jun 17, 2023 • edited Loading

jkomoros commented Jun 24, 2023

jkomoros commented Jun 24, 2023

jkomoros commented Jun 24, 2023 • edited Loading

jkomoros commented Jun 17, 2023 •

edited

Loading

jkomoros commented Jun 24, 2023 •

edited

Loading